Financial Cryptography

On googly questions -- the Dunning-Kruger effect

I suddenly came across this twitter gfx by one Ian Bremmer:

So true, and in responses a reader referred to the Dunning-Kruger effect:

The Dunning-Kruger effect is a cognitive bias manifesting in two principal ways: unskilled individuals tend to suffer from illusory superiority, mistakenly rating their ability much higher than is accurate, while highly skilled individuals tend to rate their ability lower than is accurate. In unskilled individuals, this bias is attributed to a metacognitive inability of the unskilled to recognize their ineptitude. Skilled individuals tend to underestimate their relative competence, erroneously assuming that tasks which are easy for them are also easy for others.[1]

David Dunning and Justin Kruger of Cornell University conclude, "the miscalibration of the incompetent stems from an error about the self, whereas the miscalibration of the highly competent stems from an error about others".[2]

More resonance, and more angst. A problem with the IT industry is that, because the actual measurement of the real skill level is typically too expensive, we overestimate on the basis of self-claims. It gets worse in security.

But the killer is that our entire employment system is oriented to detecting those who push through with bravado and exaggerated claims -- blagging it is called in at least one country.

What happens when the company adopts this to heart? There is research out there ("The Value of Hiring through Referrals," April 24, 2013, Burks, Cowgill, Homan, Housman) that the best employment technique (correlated with success that is) is to use your own people to recommend people like themselves. But this technique has a dark side, in that if your people are no good, they recommend ... people that are no good.

The Spence thesis is that this feedback cycle means that drift is inevitable without a correcting force. I.e., over time the blaggers take charge. Blagging becomes self-perpetuating, a sustainable cancer on your company. And thence the economy.

Once inculcated, it will be almost impossible to root out because the blagger is better at pretending to root it out as well! Of course.

So what to do? Experimentally speaking:

Dunning and Kruger proposed that, for a given skill, incompetent people will:

1. fail to recognize their own lack of skill;
2. fail to recognize genuine skill in others;
3. fail to recognize the extremity of their inadequacy;
4. recognize and acknowledge their own previous lack of skill, if they are exposed to training for that skill.[5]

That 4th point is a novel solution: if you can identify them, train them in what they don't recognise. But it does rather make some assumptions about the all-knowing experimenter.

(If this post resonances, check out the rant in 3 parts: I, II and III)

Posted by iang at November 4, 2014 09:53 AM